►
From YouTube: IETF95-6TISCH-20160404-1400
Description
6TISCH meeting session at IETF95
2016/04/04 1400
B
B
D
D
As
a
reminder,
we
remain
we're
taking
minutes
I
thank
the
minute
takers
for
taking
minutes,
so
this
would
be
Charlie
and
Dominique
and
I
think
coma
as
well.
I
don't
know
if
I
see
coma
and
then
the
meeting
is
being
recorded
presents
as
long
as
I
will
publish
all
of
that
to
the
mailing
list
right
after
the
end
of
the
the
end
of
the
meeting,
blue
sheets
are
being
distributed.
First
kind
of
you
could.
D
D
This
is
the
agenda
that
we
have
prepared
for
today,
we'll
start
by
cement
row,
which
is
what
I'm
doing
we
didn't
have
a
discussion
about
the
new
charter
and
status.
As
you
know,
we
have
weary
charted
a
couple
of
weeks
ago.
We
want
to
make
sure
that
everybody's
in
sync
and
in
what
we
did.
We
don't
have
a
very
quick
presentation
on
the
plug
test.
D
Zero
draft
01
Michael
will
then
present
the
sales
of
the
work,
an
action
plan
on
security
and
then,
if
time
permits
but
I
hope
time
time
permits,
satish
will
be
presenting
the
aodv
ripple
draft
00
here
is
there:
are
there
any
remarks
about
this
agenda?
Anything
we
want
to
change
anything.
We've
forgotten
so
I
hear
nothing
from
the
room,
but
I'd
like
to
make
a
change.
Tato.
D
E
I'm
Tarek
a
minha
princesa
core,
there's
a
draft
I'm
here
with
this
trust
given
in
a
todo
15
ie.
If
you
go
to
the
next
slide,
so
this
draft
is
because
there
have
been
just
discussing
about
in
six
slow
and
sixties
and
other
working
groups
here
to
heaven
you
run
you
know:
50-something
IETF,
protocol
direct
over
15,
4
frames
and
through
that
we
have
to
have
something
like
that
information
elements.
C
E
You
know
something
like
that:
something
information
in
their
praise.
The
15
for
has
information
elements,
there's
16
of
those
numbers,
so
they
won't
give
out
them.
Everybody
table,
give
them
one
for
each
standardization
organization,
so
60s
can't
get
one
ietf
can't
get
one.
So
if
we
make
a
document
that
will
request
them
to
give
us
a
number
and
also
you
know,
provide
a
way
to
say
that
okay,
this
is
how
we
subtype
it,
meaning
that
we
don't
need.
E
We
can
manage
with
only
the
one,
because
we
have
our
own
sometime
in
there
and
so
I
rotorcraft
about
this
saying.
Okay,
this
is
people
to
ask
this
number
now.
So
then
it
is
ready
when
I,
like
60s
or
600
ones,
to
use
this,
and
then
it
also
provides
this
very
simple.
It's
in
the
bottom
of
the
page,
this
we
have
a
one-bite
subtype
ID
there
and
that
we
have
researched
some
numbers
there
to
be
so
the
TV
everyone
out
to
be
touched.
E
You
no
use
the
subtype
idea
of
zero
and
Putin
or
16-bit
supplied
a
soup
subsiding
ID
there
in
the
extra
stuff
like
that,
and
it
also
you
know,
change
the
way
the
subtype
IDs
are
allocated
by
iono,
so
I
unavailable
allocate.
Those,
and
one
of
the
things
is
we
need
to
decide
is
how,
to
you
know,
do
this
allocation.
Do
we
want
to
heaven
you
know,
expert
review
or
Ruby.
Just
have
a
you
know:
publication
required
or
something
what
about
the
budget?
What
is
the
iona
rules
for
that?
E
E
I
already
know
that
the
coop
people
want
to
use
this
because
I
talk
to
the
cars
today
yesterday
and
he
said
yeah
I
will
get
ready
to
draft
and
what
to
use
this,
but
anyway
so
check
it
out
and
no
comment
that
I
don't
get
what
would
be
the
best
place
to
comment,
1860s
or
60,
or
what?
What?
What
list,
but
it
to
the
final
comments,
content
to
me
directly
or
two
guy
definitely
started
like
that.
I,
don't
know
any
comments,
questions.
D
E
Drunk
is
my
truck
is
going
to
be.
You
know
this
is
kind
of
framework.
It
says
that
okay,
this
is
how
the
IETF
is
going
to
be
using
this
information
of
it,
and
it
creates
one
eye
on
a
registry.
Tonight
can
I
expect
you
know
60s
if
61st
use
it
they
in
their
own
draft,
whatever
protocol
prob,
they
say
that
I'd.
E
G
E
F
G
H
E
B
D
I
So
the
first
use
we
have
is
we
finally
progress
with
the
minimal
trough,
so
it's
been
slow
because
many
things
were
happening
in
power,
so
where
we
are
now
is
we?
The
authors
have
answered
all
the
questions.
We
think
that
that
we
get
during
the
interior
review
and
we
posted
a
very
long
reply
to
suresh
Suresh.
You
show
me
and
Ralph
so
now
the
ball
is
in
your
camp
to
actually
tell
us,
maybe
you
can
recirculate
it
because
I
know
at
least
Hollis
was
not
there
unless
so
so,
Suresh,
yes
and
Ralph.
I
I
That's
the
sixth
floor,
rotten
either
draft,
and
so
this
draft
has
been
split
into
a
piece
which
we
control,
the
coding
which
is
staying
at
six
low
and
another
piece
which
is
really
the
one
that
is
dedicated
to
encode
the
ripple
artifacts
in
the
data
packets
and
six,
though,
seemed
happy
with
this
work.
So
it
is
now
moved
on
to
roll
and
both
of
those
documents
are
in
law
school.
So
I
hope
that
this
week
will
announce
that
the
last
go
our
past,
but
that's
where
they
are.
They
are
both
in.
G
Yo
yo
so
about
the
previous
others,
without
about
the
previous
slides,
yeah
I
just
wanted
to
ask
the
group
s
cultural
sick
of
ziploc
that
if
you
could
contribute
comments
on
specifically
the
paging
dispatch,
it
finished
working
good,
classical
we've
had
one
series
of
comments
recently
from
eNOS.
Thank
you
very
much,
but
it
would
be
great
to
have.
Maybe
you
know
one
or
two
more
and
so
that
we
can
cleanly
call
it
as
a
successful
working
class
comment
move
forward.
I
So
so,
please
I
actually
a
code.
What
Gabrielle
is
saying:
I
sent
a
man
on
the
mailing
list
asking
people
to
chime
in
on
this
coding
dispatch
and
six
leverage
06
low
and
respectively.
You
have
not
seen
a
lot
of
activity.
So
then,
again,
please
consider
giving
a
few
words
on
those
mailing
list
to
support
the
draft.
If
you
think
they
should
go
so
six
man
is
also
doing
things
of
interest
for
us
in
particular,
there
is
this
discussion
about
the
hop
by
hop,
which
is
getting
quite
hard
at
six
mad.
I
So
you
know
that
role
is
using
the
haba
haba
door
to
signal
repo
FBI
report
by
get
information
in
a
in
an
option
there.
So
the
discussion
is
not
at
all
whether
we
should
stop
doing
this
right
I
mean
deprecating
is
not
saying
it
doesn't
work,
but
they
are
reconsidering
preventing
anybody.
Any
new
usage
of
those
hop-by-hop
there
is
something
which
is
affecting
us
a
little
bit
closer
is
the
fact
that
the
initial
decision
for
repo
was
that
the
option
would
start
with
a
60.
I
So
there's
this
bit
on
the
left,
which
was
set
and
italy
means
that
if
you
don't
understand
how
by
up
option,
then
you
have
to
drop
it.
And
now,
since
the
discussion
at
six
man
is
a
lot
all
about
ignoring
the
hope.
I
hope,
if
you
don't
understand
it
or
you
don't
want
to
see
it,
we
would
be
interested
in
lowering
that
adoption
into
the
four
tease
so
that
this
bit
is
now
reset,
but
obviously
that
will
impact
the
existing
implementations.
I
Second,
big
news:
thanks
for
thanks
Brian:
actually
we
have
we
have
we
chartered
so
we
there
are
some
items
like
the
minimal
draft,
which
we
consider
finished
as
part
as
far
as
the
working
group
is
concerned,
and
then
there
are
some
items
which
we
differ.
Then
you
will
see
that
in
the
new
charter
and
that's
mostly
the
the
model,
the
data
model
that
that
we
plan
to
produce
for
the
60
devices
in
photograph
or
six
top.
I
I
Security,
that's
an
interesting
one!
You
know
that
we
have
discussed
the
joint
process
or
six
dish
at
length.
We've
informed
a
design
team.
At
some
point.
The
work
has
kind
of
stalled
waiting
for
the
reach
outer.
Now
we
reach
out
it.
We've
got
this
item
chartered
and
really
we
we
hope,
expect
that
the
good
work
that
had
restarted
well
at
six
days
will
now
reboot
and
Michael
will
be
presenting
that
remotely
after
as
a
deal,
the
architecture
continues.
I
Obviously
it
gives
maintaining
the
status
of
where
we
are,
and
we
also
have
this
draft,
which
is
where
advanced,
producing
our
requirements
to
death
net
for
the
60
trucks.
Now
there
are
number
of
options
for
the
tracks
and
weary
want
to
to
get
new
advice.
I'm
hearing
things
in
the
corridors
and
I
really
would
like
to
see
them
on
the
mailing
list.
In
particular.
What
kind
of
trikes
do
we
want?
I
Are
they
always
solid
tracks
that
stay
there
forever
or
are
their
needs
for
more
dynamic
track
establishments
for
just
one
shot
flow
and
then
we're
gone?
The
protocols
may
be
different.
The
approach
must
be
different.
If
you
have
this
sort
of
shoes
case,
it's
now
is
a
good
time
to
chime
in
because
we'll
be
producing
the
requirements,
and
this
has
already
started
and
then,
as
you
see,
the
data
models
are
deferred
till
we
get
more
feedback
from
the
implementations
milestones.
I
I
The
next
items
will
start
today
it's
going
to
be
putting
six
top
and
SF
0
well
getting
them
adopted
by
the
working
group
for
the
item
two
and
one
or
two
of
the
Charter,
and
then
you
see
the
the
next
items.
Oh
important
in
Berlin
will
have
the
third
block
test
that
would
be
presented
later
on
during
this
session
as
well.
That's
that's
a
huge
event
for
us.
So
if
you
have
a
chance,
if
you
have
an
implementation
and
you
have
a
chance
to
participate,
it's
free,
there
is
a
fantastic
set
up.
I
I
Ard
action
plan,
basically
for
the
future
loader.
The
first
thing
is,
you
know,
keep
this
child
development
phase
rolling
with
the
HC
black
test,
so
we'll
see
what
we
have
in
in
plan
for
for
the
black
test
in
Berlin
and
and
kickstart.
The
joint
process,
the
SF
0
workers-
is
very
well
started.
The
sixth
up
work
is
very
well
started.
The
tracks
is
very
well
started.
D
Okay,
so
Megan
are
he'll,
put
together
these
slides,
but
couldn't
come
so
he
asks
me
to
present
it
on
his
behalf.
I
hope
I'm
not
saying
too
many
stupid
things.
So
this
is
a
report
about
the
second
success
block
test
that
was
organized
in
February
in
Paris,
so
there's
two
pieces
to
this
presentation
very
quick
overview
of
what
we
did
and
then
very
quick
introduction
of
the
third
block
test
in
Berlin.
D
So
the
event
there
were
12
participating
companies
and
five
observer
companies.
We
had
test
session
that
lasted
for
two
hours
and
we
basically
went
through
a
combination
of
different
companies
face
to
face
companies
to
to
test
different
pieces
of
a
test
plan.
This
is
the
agenda.
You
can
look
at
it
on
the
slides,
I,
don't
want
to
spend
too
much
time
on
it.
D
There
was
we
use
a
tool
that
allowed
us
to
record
the
outcome
of
the
of
the
tests,
and
these
are
the
tests,
so
we
had
tests
on
synchronization.
We
had
tests
on
on
ripple
a
minimal
test
on
triple.
This
was
not
testing
complete
rebel
but
just
making
sure
that,
in
the
context
of
six
ish
ripple
was
implemented
correctly.
Tests
on
the
sixth
of
protocol,
as
well
as
tests
on
60
rh,
which
are
here
so
the
to
the
main
point
of
this
blog
test,
was
the
6th
or
protocol
and
six
lower
age.
D
These
are
the
the
outcome
and
you
can
see
that
the
outcome
is
a
hundred
percent
rely.
Is
one
hundred
percent
interoperability
now
take
this
with
a
grain
of
sand?
People
don't
like
to
report
when
it
doesn't
work
so
so
you
meet
with
it.
You
meet
with
a
little
company.
What
doesn't
work?
You
say
like?
Oh
well,
we
didn't
test
it
when
it
works,
you're
very
happy
to
press
the
button.
Yes,
it
didn't
work
so
there's
a
hundred
percent
reliability,
but
interoperability.
C
D
But
the
the
actual
number
you
need
to
look
at
is
sixty
percent
coverage,
so
we
were
able
to
do
sixty
percent
of
all
the
combinations
of
all
the
companies
all
the
tests
during
during
a
protest.
It
was
just
a
two
day
event.
So
there's
no
way
we
get
to
a
hundred
percent,
but
were
you
know
pretty
pretty
pretty
happy
to
get
to
to
to
sixty
percent?
We
gave
out
a
gold
device.
D
This
is
an
open
mode
over
a
pre-programmed
with
the
open,
wsn
implementation
that
served
as
a
reference
implementation
that
we
sent
to
every
participant,
allowing
the
participant
test
before
coming
to
the
event,
and
this
allowed
basically
the
event
to
run
very,
very
smoothly.
We
had
in
that
we
didn't
have
any
big
big
surprises
therapy.
Thanks
to
this
goal
device,
you
can
look
at
the
slides
for
having
the
specific
test
results.
D
If
you
want
to
see
like
test
by
test
and
the
conclusions
that
we
we
were
able
to
test
six
basic
66
stop
protocol
implementation,
allottee,
so
we
tested
add,
delete
all
the
little
older
little
commands.
We
did
not
test
scheduling,
function,
zero,
which
is
the
algorithm
that
sits
on
top
of
the
protocol,
because
we
that
we
consider
this
product
that
this
work
was
not
advanced
enough
to
go
into
a
into
an
intro
prevent.
D
We
did
some
six
lower
age
demonstrations
but
didn't
really
get
to
the
actual
testing
of
six
lower
age,
because
the
sex
or
age
standard
dress
had
likes
stabilized
only
a
couple
of
weeks
before
so
not
everybody
had
it
had
it
had
it
implemented.
So
we're
going
to
ask,
as
per
the
plan
of
last
year,
we're
going
to
continue
doing
in
this
series,
so
we
had
the
first
event
and
prague
focusing
on
minimal
and
ripple.
D
We
had
the
second
event
in
Paris,
focusing
on
minimal
as
well,
but
documented
with
six
top
protocol
and
a
bit
of
six
or
h
and
in
Berlin
we
still
have
to
define
specific
view,
scope,
I.
It's
not
it's
not
a
promise,
but
it
will
probably
be
continuing
on
six
or
a
chance
text
of
protocol
and
probably
adding
scheduling
function,
maybe
some
backbone
router
functionalities.
So
it's
it's
a
it's
a
you
know,
it's
very
high!
Sorry,
it's
it!
D
We
will
have
the
event
we
still
have
to
put
together
these
female
experts
and
decide
on
the
scope
of
this
event.
But
let
me
let
me
highlight
how
you
know
how
cool
this
after
the
July
event
will
have
fully
functional
implementations.
You
know
implementing
the
fool
set
of
specs
of
six
dish,
which
is
which
is
which
is
a
good
milestone
for
us
question.
Let's.
I
Get
you
better
tomorrow,
we
discussed
the
backbone
router
as
well
for
growing
is
not
written
there.
I
know
some
implementers
want
to
come
and
try
without
booking
their
registration
right
now.
Yeah.
D
Yeah
say
so:
my
apologies
for
I
said
it
orally.
It's
not
on
the
slides
something
some
communication
didn't
happen,
but
absolutely
backbone.
Router
is,
is
a
very
high
on
the
list
of
things
that
will
be
considered
for
the
photo
July
meeting
for
the
July
podcast,
and
so
this
would
be
16
17
right
before
the
IDF
meeting
in
Berlin
yeah.
D
D
J
J
D
J
D
D
J
Sorry
for
that,
so
what
we
did,
it
will
rename
several
fields
on
the
on
the
packet
format
and
on
the
on
the
information
element
that
was
container
and
we
rename
it
to
metadata.
Then
we
rename
the
term
talking
to
sequence
number
and
then
we
add
another
another
transaction
which
is
the
three-way
three-way
them
in
transaction
that
I
will
describe
later.
Can
you
go
to
the
next
slide
please?
J
So
the
basic
idea
is-
and
this
in
this
protocol
is
to
enable
the
negotiation
of
cells
between
pairing
notes.
So
the
protocol
defines
messages
between
two
nodes
in
the
selfish
network
in
order
to
add
cells,
the
cells
or
get
information
or
count
the
number
of
cells
etcetera
that
we
have
in
the
level
of
these
two
neighbors
Nexus
life.
J
J
What
about
now
better?
Okay,
next
slide,
please.
J
So
then,
the
draft
this
explains
how
to
use
six
top
in
in
together
with
the
minimal
draft,
basically,
is
that
we
use
another
a
slot
frame
to
allocate
the
resources
and
minimal
runs
in
a
slot
frame
0,
while
six
top
uses
lot
frame
1-
and
this
is
not
a
mandatory-
so
it's
something
optional
next
slide.
Please.
J
Okay,
so
this
is
the
the
transaction
that
we
defined
is
a
two
steps.
Transaction
requesting
node
sends
a
request
message
to
the
counterpart,
indicating
how
many
cells
it
once
to
locate
and
also
a
list
of
candidate
cells,
and
then
the
receiving
node
confirms
the
the
allocation
with
a
response
message
and
indicates
which
of
the
cells
in
the
cell
list
are
finally
used
next
slide,
please
so
that
the
3-step
transaction
works
in
another
way.
J
It
is
the
requester
note
tells
only
how
many
it
once
the
receiver
note
returns
back
a
response
with
a
candidate
list,
and
then
the
requester
node
sends
sends
back
the
confirmation
confirming
which
Lee,
which
of
the
cells
in
the
least
our
are
finally
allocated
it's
a
three-step
process
next
slide.
Please.
J
J
D
K
D
K
J
J
J
And
these
are
the
two:
let's
say
types
of
messages.
One
is
for
add
delete
where
we
specify
the
number
of
cells
that
we
want
to
allocate,
and
then
the
cell
list
and
the
metadata
field,
which
is
used
by
the
scheduling
function,
or
example,
to
say
if
these
cells
have
to
be
allocated
even
specific,
a
slot
frame
or
if
they
have
to
be
allocated
in
a
in
a
chunk
or
any
other
rule,
that
the
scheduling
function
once
wants
to
define.
So
for
for
the
six
top
protocol
disease
transparent.
J
J
So
then,
for
the
response
message:
format:
we
have
the
same
format
as
a
response
on
in
the
case.
That
is
it
it's
an
add,
delete
or
at
least
the
response
returns.
The
cell
list,
as
I
said
before
in
case
of
a
count,
it
returns
the
number
of
of
elements
in
the
schedule
or
it
if
it's
a
clear,
it
doesn't
return
anything
in
the
other
fields,
part
the
next
slide.
J
J
J
So
this
is
the
process
that
we
defined
before
four
for
the
two-way
transaction.
We
have
the
node
deciding
the
scheduling
function,
deciding
how
many
cells
to
add
and
then
preparing
the
candidate
cell
list
and
and
sending
another
request
to
the
sixth
of
protocol,
which
sends
the
packet
over
the
air
to
the
other
node
and
then,
after
some
verification
and
trying
to
see
there's
a
pre-sales
for
for
the
requested
cells
in
the
response
schedule.
The
note
the
supernote
responses
with
with
the
result
of
the
function
next
slide
I'm.
J
J
So
the
draft
also
adds
these
guidelines
for
the
scheduling
function.
It
adds
a
kind
of
proposed
instructor
for
the
document
and
also
defines
the
range
of
identifiers
having
most
of
them,
managed
by
us
and
living
few
of
them
and
managed
and
to
be
used
by
different
scheduling
functions
next
slide.
Please.
J
D
D
It
has
been
the
kind
of
the
main
objective
of
one
of
the
pluck
tests
and
in
terms
of
implementation,
if
I
count
correctly,
there's
it's
fully
implemented
at
the
open
areas
and
implementation,
and
there
are
three
other
open
source
of
limitations
and
who
I
believe
commercial
implementation
of
this.
So
it's
been
through
through
quite
a
bit
of
the.
D
D
L
L
L
So
the
idea
of
these
trophies
to
define
us
getting
in
function
according
to
the
requirements
from
this
a
60
protocol
draft-
and
we
have
some
news:
I-
have
changed
some
relevant
things
and,
first
of
all,
we
have
changed
the
bandwidth
estimation
algorithm,
since
we
are
not
supposed
to
now
that
they
scheduled
a
be
the
application.
Hatha
has
to
request
bandwidth
to
the
lower
layers.
L
We
had
a
bit
fresh
differentiation
between
bandwidth
and
cells
because
there's
not
not
a
similar
thing
and
we
have
added
the
whitelist
and
blacklist
I
idea,
which
was
expressing
one
of
the
interim
meetings,
how
long
to
do
list
and
Ian.
So
the
one
with
dementia
algorithm
is
basically
on
the
mission,
the
availability,
the
neighbor
bound
with
requests
and
local
bad
with
requests.
Okay,
so
we
have
to
talk
I
political
from
both
items.
So
before
we
had,
we
have
thought
that
we
had
to
establish
some
requests
and
this
assumption
from
the
application
does
not
apply
anymore.
L
So
we
have
built
this
new
algorithm.
We
have
to
estimate
our
current
bandwidth
or
what
we
are
we
to
use
from
the
currently
used
bandwidth.
So
we
have
the
first
part
on
the
left,
which
is
directly
what
what
we
are
doing
about
doing
before.
We
are
just
using
the
request
that
we
had,
and
on
the
other
side,
we
just
added
a
little
bit
more
of
work
on
just
what
some
extra
and
some
over-provisioning
on
what
should
be
the
real
done
with
we
are
using
this
way.
L
You
have
indirectly
estimated
indirect
instead
to
measure
of
the
local
bandwidth
requirements,
and
since
we
are,
we
wish
to
compensate
the
underestimated
bandwidth.
We
shall
get
from
the
application,
so
we
have
another
alternative
we
have
been
discussing
these
days
being
and
between
when
the
moment
way,
we
submitted
a
draft
and
now
and
that
the
right
part
that
we
had
before,
as
you
can
see
in
this
part,
we
can
replace
the
directly
to
this
arrangement
with
the
allocation
policy
changing
this
sets
your
threshold
that
we
used
before,
instead
of
directly
on
the
side.
L
Then
we
are
going,
we
have
considered
their
relocation
request
now:
okay,
yes,
there
are
an
incoming
bandwidth
requirement.
Okay,
that's
a
new
idea
that
meaning
that
any
changes
may
change,
also
the
requirement
under
requirement
so
that
we
are
expected
to
increase
the
packet
delivery
rate
whenever
you
have
to
have
a
relocation
request
and
then
I
am
a
bandwidth
related
to
a
number
of
cells,
of
course,
and
itself,
a
racket,
liberal
right.
That
makes
a
difference
between
the
the
original
bandwidth
requirement
with
to
the
number
of
cells.
L
So
we
have
to
relate
it
through
a
number
of
like
a
question
or
a
constant
value.
It
that
depends
on
each
of
the
cells.
On
the
other
side,
we
have
worked
out
the
whitelist
and
blacklist
mechanism,
meaning
that
on
the
tool
and
to
transaction
authorized
end
section,
and
so
you
see
to
pakistan
three
pakka
transaction.
Sorry,
we
have
kept
the
whitelist
as
the
one
before
did
we
change
bad,
we
added
one
with
a
blacklist.
L
The
meta
data
is
only
fair
to
define
the
the
wine
list
or
a
black
list.
Oh
yeah,
we
are
reserving
it.
The
question
is
what
luba
the
reserve
bits?
Okay,
we
have
reserved
only
toes
and,
of
course,
will
be
identifying
this
load
frame
here
and
they
know
behavioral
boot.
We
have
this
proposal
by
a
scale
that
instead,
okay,
maybe
we
should
use
a
limited
pool
of
cells
to
accelerate
the
joint
project
I'm
process
and
then
return
this.
L
M
L
I
C
I
L
I
I
remember
at
some
point:
we
had
that
discussion
that
if
minimal
is
there
anyway
yeah,
then
we
could
have
enough
minimal
sales
where
everybody
is
listening
to
that.
If
you
get
a
berth
for
any
individual
child,
then
you
would
put
his
burst
in
the
minimal
and
and
you
could
decide
in
SF
0.
Oh
this
guy
is
talking
a
lot
on
the
minimal.
Let's
give
him
more.
I
L
What
we
have
to
think
about
that
we
did
lots.
Of
course
you
did
thank
you.
N
D
D
L
D
K
D
So,
just
as
a
as
a
context,
so
this
is
SF.
0
scheduling,
function,
0
there
are
256
numbers
for
identifying
scheduling,
so
the
idea
of
SF
0
is
that
it
is
the
simplest
and
a
more
generic
one.
We
imagine
that
people
that
it
will
not
you
know
kind
of
almost
by
definition,
will
not
satisfy
all
needs,
and
if
there
are
some
other
needs
that
that
this
would
become
sf1
SF
to
SF
3
or
become
an
SF
in
a
range
of
those
unmanaged
or
vendor
specific
implementations,
which
is
also
possible.
C
D
So
I
think
it's
we
do
a
preliminary
co-production
we
might
might
have
raised,
raised
hands
yeah.
So
these
these
two
drafts
are
our
companion
drafts
right.
So
the
sixth
of
protocol
draft
is
the
protocol,
and
the
discussion
function
is
the
the
policy,
the
algorithm
that
uses
that
that
sits
on
top
of
the
program
that
drives
it.
These
two
things
have
been
worked
on
for
a
number
of
months.
Now
they
are
only
just
now
part
of
the
Charter.
D
I
I
C
I
Ok,
well
that
still
positive
constants
s,
plus
whether
the
jabber
okay.
So
let's
confirm
that
on
the
mailing
list
as
well
and
if
you
did
not
raise
hand
if
you
want
to
contact
us
with
a
good
with
your
reasons,
I
mean
real
estate.
Okay.
So
what
time
being
we
think
that
the
group
is
favorable
to
the
adoption
of
both
documents
and
we'll
compare
that
other
mailing
list.
D
A
D
N
D
M
O
You
thanks
to
my
name
yep
there
we
go
perfect.
Now
you
can
hear
me.
P
Go
ahead,
Mon
again
all
right,
I'm
on
it,
since
I'm
published
next
I
caught
next
slide.
Please
look
at
the
camera
I
like
the
slide.
Oh
okay.
So
the
summary
of
the
summary
is
from
the
six
dish
point
of
view.
Not
a
lot
has
changed.
We
went
through
a
bunch
of
simplifying
things
last
time
about
why
the
six
dish
is
a
bit
different
than
some
other
networks.
One
of
them
is
that
we
assume
that
we
have
some
kind
of
management
and
infrastructure.
That's
going
to
help
us
out
arm.
P
P
The
other
simplifying
assumption
between
made
some
time
ago
was
that
there
would
be
manufactured
installed
certificates
of
some
kind.
Some
people
would
like
to
have
shared
keys
of
various
kinds,
and
we
propose
that
that
we
come
back
to
that
as
a
different
version
of
this,
that
we
don't
think
that
this
scales
to
the
large
at
all
on
and
I
think
it
will
be,
maybe
a
distraction.
P
So
so
a
lot
of
the
work
and
a
lot
of
the
work
has
moved
out
of
the
six
dish
working
group
into
the
animal
working
group
specifically
into
the
bootstrap
design
team,
which
consists
of
myself
max
Pritikin
peter
van
der
stoke.
We
have
robert
craigy
joins
us
regularly
and
carless
and
Michael
Berenger
as
well.
P
So
a
little
bit
of
contrast
between
anima
and
sixth
ish
animas
goal
is
to
get
a
some
kind
of
secure
connection
established
with
the
registrar.
Within
that
secure
connection,
we
will
run
a
protocol
called
enrollment
over
secure
transport,
basically
as
a
TLS
type
connection
in
which
you
do
some
rest
full
operations
and
get
a
certificate.
On
the
other
hand,
the
goal
of
the
six
dish
security
is
to
bring
up
a
six-top
connection
to
the
jce
and
pce
in
which
we're
going
to
run
six
top
to
actually
do
provisioning
our
configuration.
P
So
this
is
a
slight
difference
there
as
to
what's
going
on,
anima
accommodates
https
and
it's
been
a
bit
of
a
hard
sell
to
say
the
dtls
should
be
the
mandatory
to
implement,
but
perhaps
that
will
get
resolved
we're
a
sixth
ish
would
prefer
to
do.
Dtls,
co-op,
plus
six
top
only
I,
don't
know
whether
block-wise
mode
of
co-op
would
be
controversial
or
not
within
six
dish,
but
that
is
probably
the
way
to
get
around
the
potentially
the
MTU
issues
of
a
larger
objects:
the
sixth
ish
stuff.
P
P
We
assume
that
our
devices
are
constrained
in
RAM
and
that
they're
battery-powered
anima
network
is
not
really
constrained,
but,
as
it's
brought
up,
it
might
be
as
constrained
as
a
six
stitch
network.
In
fact,
it
may
even
be
running
something
like
15
dot,
4
to
bootstrap
itself,
but
once
it's
running,
maybe
turning
on
multi
gigabit,
fiber
pipes,
and
so
it
can
do
a
lot
of
things
later
on
that
that
sixth
ish
can't
do
so.
That's
some
11
there.
P
So,
as
I
said,
while
the
well
the
machine
that
the
big
router
that
you
know
you
don't
want
to
drop
on
your
toe
because
you'll
wind
up
the
hospital
itself
is
not
constrained.
The
line
cards
and
the
CPUs
on
those
line
cards
may
be
as
constrained
to
some
of
the
devices
that
we
work
with.
So
some
consideration
to
think
about
its
commonality.
Next
slide,
please.
P
So
all
my
greenboxes,
the
wrong
spot,
that's
okay,
anima
and
net
cough
there's.
Some
interesting
contrasts.
P
Anima
is
interested
in
secure
transport,
whereas
net
cough
is
interested
in
yang
and
there's
some
other
differences
here.
Animas
morgan,
boat,
bringing
up
infrastructure
devices
and
net
cough
is
sorta
more
about
bringing
up
telephones
and
home
routers
and
things
that
are.
There
are
many
many
many
of
them
and
they're
relatively
similar.
This
slide
is
not
talk
about
six
dish.
Let's
continue
on
next
slide.
P
So,
sixth
ish
versus
netcom,
so
some
of
the
things
that
really
really
connect
us
is
we're
both
really
interested
in
yang
type
things
crossing
over
our
secure
Channel.
So
that's
something
to
think
about
that.
There's,
a
very
strong
relationship!
Sixth
ish
wants
to
replace
its
I
dev
ID
the
manufactured
install
certificate
with
some
other
local
domain
ID
ID-
and
we
need
this.
P
If
we're
going
to
use
that
that
ID
to
do
anything
like
80
to
59
a
key
management
protocol
on
a
per
link
layer,
whereas
net
comp
well,
it
can
replace
the
I
dev
ID
with
an
L
divided
eithe
ATS,
a
not
a
built-in
must
do
kind
of
thing.
It
kind
of
it
assumes
that
the
idea
of
ID
record
hangs
around
a
lot
longer
until
it's
it's
connected
to
its
real
network.
Next
slide.
Please.
P
So
what
fundamentally
is
the
problem?
Maybe
I
should
have
put
the
slide
first,
it's
basically
how
to
how
to
add
random
nodes
and
assume
that
they
might
be
malicious
and
that
they
want
to
add
them
without
destroying
your
network
and
there's
a
bunch
of
things
that
that
you
can
do
that.
This
way.
P
For
instance,
one
of
the
issues
with
with
with
some
of
the
EAP
pana
solutions
which
which
have
been
discussed
on
and
off
in
the
sixth
ish
security
group,
is
that
they're
they
can
be
big
a
bunch
of
code
that
gets
used
only
once
we're
not
really
that
fond
of
having
TLS
can
EAP
code
in
our
small
devices,
which
we're
not
really
going
to
ever
use
a
gun
and
there's
also
even
a
concern
that
in
devices
that
are
not
six-stage
devices
that
the
One
X
function
in
the
device
might
you
might
actually
need
it
for
authenticating
end
end
hosts
on,
and
that
may
be
confusing.
P
So
this
is
a
diagram
kind
of
shows
that
the
contrast
between
the
two
methods,
the
key
thing,
is
that
they
both
involve
manufactured
install
certificates
in
some
kind
of
supply
chain
process
by
which
you're
going
to
get
some
kind
of
ownership
voucher,
which
allows
you
to
tell
the
end
node
that
in
fact
it
belongs
to
your
network
rather
than
a
different
network.
Next
slide,
please
and
again,
anima
versus
net
cough
is
a
little
bit
different
again,
but
again
it's
the
same
supply
chain
and
vendor
installed.
Certificates
next
slide.
P
So
one
of
the
key
problems
that
we've
discussed
is
how
to
proxy
or
tunnel
traffic
from
a
new
node,
which
has
shown
as
the
little
thing
with
the
wheels
on,
is
trying
to
go
through
a
joint
assistant
and
is
trying
to
get
to
a
registrar.
And
so
six
dish
is
pretty
good
at
building
those
green
pipes
as
as
tracks.
P
And
so
we
could
build
quite
easily
build
a
track
that
goes
up
to
the
registrar
and
we
could
limit
how
much
bandwidth
is
used
and
that's
a
really
important
way
of
how
we're
going
to
restrict
the
ability
to
dial
service
in
there.
One
of
the
concerns
that
came
up
with
doing
straight
and
capsulation
of
link
layer
packets
up
to
the
registrar,
is
what,
if
the
joining
assistant
has
multiple
interfaces,
and
the
answer
is
actually
surprisingly
simple.
It's
not
to
use
the
ula
on
that
thing,
but
rather
to
recognize
that
the
joint
assistant
can
have.
P
We
have
more
than
one
v6
source
address.
So,
given
that
it
can
accommodate
more
than
those,
it
can
quite
easily
figure
out
which,
when
the
packet
returns,
which
link
it's
supposed
to
root,
send
the
packet
back
to
so
that
solves
a
problem
that
have
been
bothering
several
people
for
some
days
next
slide.
Please.
P
So
there's
a
bunch
of
different
ways:
we
can
do
this,
and
we've
discussed
this
at
length
in
anima.
We're
going
to
discuss
it
again
for
this
afternoon,
I'm
not
going
to
go
a
great
deal
into
this
things
like
NAT
66
were
discussed,
circuit
layer,
proxies
or
HTTP
proxies,
which
would
all
involve
huge
amount
of
state
and
RAM
in
a
joint
assistant
devices
and
I.
P
D
D
Do
you
believe
so
you
started
the
presentation
with
the
with
the
indication
that
we
started
a
lot
of
work
at
the
success
security
design
team
that
in
verse,
42
other
working
group
site,
which
I
think
is
a
great
thing,
because
because
those
things
are
way
more
generic
and
way
more
applicable
to
IOT
than
just
six
dish.
Do
you
do
you
believe
that
as
a
workgroup,
we
should
we
should
remain
on
the
sidelines.
P
The
answer
is
half
of
the
answers
yes
and
half
answer's.
No,
we
should
leverage.
We
should
leverage
the
secure
trust
relationships
that
anima
is
building
between
new
nodes
and
registrar's
or
JC
es.
We
should
leverage
the
protocol
the
encapsulation
protocol,
such
as,
essentially,
we
could
have,
on
the
network
side
of
things,
a
very,
very
similar
requirements
in
the
registrar
to
what
anima
is
doing
with
in
that
tunnel.
That's
created,
I
believe
we
should
be
running
six
top
rather
than
EST.
P
Okay,
so
that's
mostly
I
was
at
one
point
in
part
to
the
minimal
draft
and
I
believe
was
removed
again
because
it
didn't
make
sense
in
the
minimal
draft,
and
so
we
need
to
continue
to
define
our
security
bootstrap
parameters
that
live
in
the
six
top
stuff.
So
that
means
yang
definitions
and
all
the
other
things
that
we
need
to
do
to
make
that
happen
on
the
whole
k.
1
k,
2
discussion
is
is
relevant,
I
hope
is
settled,
but
it
is
it
does.
P
D
So
that
that's
exactly
where
I'm
getting
at
so
we
used
to
you
start
a
node
node.
You
join
joining
load
users
joining
it
proxy
to
go
through
a
jonas
system
to
go
through
a
jce
and
through
that
tunnel.
Once
the
security
session
is
established,
the
jce
then
provisions
in
the
very,
very
smaller
space
k
to
the
key
k
2.
D
P
P
I
P
My
my
feeling
at
this
point
is,
is
that
we
should
have
writing
code
and
we
should
document
what
happens.
That's
that's
that's
a
little
bit
of
where
I
think
we
have
to
validate
that
it
works.
We
can
see
that
we
got
all
the
right,
the
right
pieces
to
actually
make
it
happen.
Could
could
we
make
use
of
additional
authors
on
additional
documents?
P
D
D
D
C
D
K
D
D
F
F
F
F
F
F
F
F
F
We're
all
if
all
the
links
along
the
two
directions
along
the
path
from
the
source
to
destination
a
symmetric,
the
a
B
who
will
remain
to
be
0
and
for
this
case
the
destination
will
I
unicast
the
control
message
panel
to
the
sauce.
Then
the
sauce
will
get
to
know
and
the
route
the
route
is
set
up
from
the
definition
to
the
sauce.
F
Well,
it's
possible
answer
the
intermedia
nude.
We
helped
the
sales
of
available
for
only
one
direction.
In
this
case,
that
means
the
other
version
do
not
have
the
available
sales
for
the
two
to
meet.
The
requirement
of
the
route
is
correct.
Then
this
node
has
to
set
the
a
bit.
F
F
Negotiation
procedure
will
happen
to
the
cyst
top
layer
and
the
sales
will
be
scheduled
for
the
20-point
route
between
the
source
and
the
definition
in
in
this
matter.
We
do
not
use
the
address
vector
in
the
control
message,
so
for
this
reason
we
can
see
some
bandwidth
overhead
when
the
control
messages
are
transmitted
on
the
links.
I
Get
there
first
disclaimer.
We
are
internet
area
here
and
we
are
not
defining
what
protocols
per
se.
We
are
using
them
right.
So
since
you
mentioned
that
it's
aodv
ripple
suddenly
that
there
must
be
a
time
where
this
is
presented
at
roll
and
roll
adopts
this
work
and
then
six
dish
will
know
inner
it
or
work
on
it.
I
mean
it's
not
like
in.
We
cannot
be
the
group
telling
you.
Yes,
we
have
this.
We
do
this
work
in
this
room
because
we
are
not
the
writing
specialist
here.
I
Oh
change,
hats
back
home,
so
there's
just
a
disclaimer
right
I
mean
this
work
must
happen
at
all
Impala
or
if
you
issue,
because
there
is
this
resemblance
with
ripple
or
p2p
repo
as
well.
So
that's
actually
my
question
is
you
seem
to
be
doing
what
they
love
this
year
either
very
happy
with
is
but
I
see
that
your
your,
how
probably
doing
the
the
standard
track
version
of
what
p2p
people
try
to
be,
which
is
in
a
bit
of
a
ripple,
is
experimental.
I
N
I
Rephrasing
I'm
asking
we've
been
doing
an
experimental
reactive
protocol
based
on
the
iOS.
Now
probably
the
next
phase
is
certain
need
to
go
for
a
set
of
truck,
and
we
need
to
explain
what
what's
different
I
mean,
how
why
you
know
what's
new
for
people
who
know
the
previous?
If
all
well,
we
are
taking
well.
N
Roughly
speaking,
what's
new,
is
they
were
trying
to
do
it
without
the
address
vectors
and
also,
as
far
as
moving
from
the
oh,
my
name,
Charles
Newark,
and
by
the
way,
and
as
far
as
moving
from
experimental
to
proposed
standard?
Well,
presumably,
the
experimental
document
has
been
around
people
understand
it
better
now,
and
it
is
too.
We
also
like
to
have
implementation,
so
we
can
try
to
get
that
done,
but
in
the
asymmetric
nature
I
think
is
well
that
definitely
worth
consideration
to
f
in
a
single
protocol.
D
D
O
M
M
R
R
D
A
have
fine,
Tomas,
Medina
I
have
a
quick
clarifying
question.
You
address
a
lot
asymmetric
links,
and
so
you
have
a
if
I
understand
correctly.
You
I
reviewed
the
draft
like
tonight
than
I
sent
you
but
I
haven't
had
time
to
read
written
responses,
yet
I
I
apologize,
but
the
1.1
vain
words
that
you
look
at
asymmetric
links
so
basically
have
a
path
that
goes
from
A
to
B
and
then
the
multipath
absorb
e2a.
But
this
is
possibly
a
different
one
because
they're
asymmetric
yes,
what
exactly.
C
D
F
I
setting
a
and
the
campus
speak
to
each
other
right
on
the
different
directions
they
make
I
with
different
this
number
of
sales
yeah.
They
hope
to
have
the
ability
to
speak
with
each
other,
otherwise
it
they
will
not
receive
any
feedback
and
the
route
Kanal
be
discovered
successfully.
So
it's
a
asymmetry.